backup-restore-runbook-generator

Use pg_dump with custom format and compression, then upload the output file to S3 with aws s3 cp using the STANDARD_IA storage class. The generated script also verifies the upload, cleans up local backups older than 7 days, and sends a Slack notification.

Point-in-time recovery restores a base backup with pg_basebackup, then configures recovery.conf with a restore_command fetching WAL files from S3 and a recovery_target_time. PostgreSQL replays transaction logs until the target time and then promotes the instance.

A DR runbook should include incident assessment steps, recovery point identification, environment preparation, restore execution, validation, cutover, and post-recovery monitoring. It also defines RTO/RPO targets, team role assignments, and an escalation path.

Yes, the Skill provides a MySQL backup script using mysqldump with --single-transaction for consistent backups without locking tables, compressed with gzip and uploaded to S3. Restore and PITR procedures are primarily documented for PostgreSQL.

Run validation checks comparing table counts, row counts on critical tables, foreign key constraints, and index counts against expected values. The validation script also times a sample query to detect missing indexes after restore.

The Skill recommends quarterly DR drills to test restore procedures regularly. Backup health should be monitored continuously with automated alerts when no backup completes within 24 hours or when backup sizes appear suspiciously small.